AI and Vibe Coding Are Radically Impacting Senior Devs in Code Review
AI agents, vibe coding, AI code review, and other AI-centric topics are all I see developers talk about on social platforms these days. Development teams have transitioned from being skeptical about introducing AI into their codebases to controlled use of AI within their codebases, and in some cases, full-scale adoption.
Experts predict that 90% of code will be AI-generated within the next year. While this will further accelerate the software development process, it raises concerns about code quality, primarily due to the increasing prevalence of AI-driven code reviews. We now have numerous AI tools available that automate code review workflows.
- If AI generates the code and AI reviews it, how do we ensure quality
- Should AI sign off on PRs
- Should we leave the code review altogether to AI
- What will then be the role of senior developers in code review
This article will address these questions by explaining the evolving role of senior developers because of the emergence of AI-powered code review.
Limitations of Traditional Code Review
Senior developers are the traditional guardians of code quality due to their role in code review, carefully reviewing pull requests (PRs) to ensure they meet the company’s pass threshold. This process typically involves multiple back-and-forth exchanges between the PR author and the reviewer. Many say that it is a tedious process. However, it has helped development teams ensure code quality over the years. So it’s hard to disagree that it works, but it wasn’t very efficient, so static code analysis tools (e.g. linters) were created.
Static analyzers helped solve the inconsistencies and human errors in manual reviews. They helped developers analyze code automatically to detect potential issues and enforce set guidelines. The challenge with static analyzers is that they are not intelligent; they work strictly based on predefined rules and do not understand the intent behind the code.
Let’s Examine These Limitations in More Detail
Focus on Repetitive and Low-level Tasks
Senior developers often find themselves bogged down in low-level code review tasks, such as style and basic error checks. While these are necessary, the time spent on them could be better utilized handling higher-level tasks, such as architectural decisions. Although linters can automate basic checks, they do not understand the context of certain code decisions.
The focus on low-level tasks can affect the quality of contributions from senior developers. “Just hire a junior to focus on these low-level tasks,” you might say. While that is an ideal solution, it is not viable for startups with limited resources.
Time-Intensive Processes
Large PRs require great diligence when reviewing. A reviewer must meticulously inspect each file, review the code, and provide detailed feedback, all of which require a significant time commitment. This can become a bottleneck for development teams with tight deadlines. However, breaking large code changes into smaller pull requests (PRs) can help speed up the review process.
Subjectivity in Reviews
Human error is a common occurrence in manual code reviews.
As a senior developer, you might sign off on a PR for approval, but overlook some errors. We are humans, so we can make mistakes or have subjective opinions. This subjectivity can cause friction within development teams, as what is acceptable to one might be problematic to another. Establishing coding standards and setting a pass threshold helps, but they do not entirely eradicate subjectivity and personal preferences in code review.
Development teams can overcome some of the limitations in traditional code review by leveraging AI agents. Don’t get me wrong, I am not saying AI is the secret elixir for all development issues, but it offers a more efficient approach to code review.
How Is AI Reshaping Code Review?
When discussing the use of AI in software development, particularly in code review, some people approach it from what I call “the angle of substitution,” implying that AI can entirely replace human reviewers. This perspective is incorrect. Any experienced developer knows that AI will not replace human judgment in code review. This is evident in the vibe coding mess we have seen online recently, where less experienced developers entirely depend on AI to write and review code.
It’s good practice to use AI as an assistant, not as a replacement for human judgment, because:
- AI lacks team-specific context. AI wouldn’t know that your team has decided to avoid a specific design approach after your last roadmap update.
- Certain types of knowledge can only be gained through years of practice.
- You cannot hold AI accountable for any issue that may occur.
- AI needs human guidance to achieve the best results.
Let’s discuss how AI can reshape the code review process.
Tasks Automation
You no longer have to spend much time on repetitive tasks. You can intelligently automate tasks such as style checks, PR summaries, and bad code detection while you verify and approve suggested changes. As a senior developer, this will allow you to be more productive and focus on more important concerns.
First Level Issues Detection
AI agents can scan code for potential issues and check test coverage. It can also offer assistance on how to resolve these issues. This can serve as a first-level check for your PRs, reducing the workload of the human reviewer.
Chat With Your Codebase
Your codebase can become complex to understand as it grows. If your organization has a large codebase, it may take new members some time to study and understand it, making meaningful contributions. Similarly, a large pull request is complex to review — the human reviewer must review each file to understand what the code does. With AI, you can get summaries for your PRs and chat with your codebase.
A new teammate doesn’t have to wait for a senior to guide them through the codebase before they start making contributions. They can discuss this with AI. This better uses everyone’s time and enables the team to build faster.
Consistency in Your Codebase
Human biases can affect code consistency. It is natural for humans to have preferences, especially senior developers. This is why enforcing a pull request (PR) benchmark is necessary for consistency. AI can ensure consistency by intelligently reviewing code against your set benchmark.
Collaborative Code Review
AI can collaborate with a senior developer to review PRs for a better and faster result. In this type of collaboration, error oversight is less likely to occur, as AI will catch any errors the developer misses and vice versa.
If you are looking for an AI-powered code review tool to get started with, I’ll recommend CodeRabbit. It offers advanced code analysis, PR summary, issues detection, and other features that enable a seamless code review workflow. It also brings AI-powered code reviews directly into VS Code, Cursor, and Windsurf.
The Evolving Role of Senior Developers
The emergence of AI as a collaborative assistant means that the role of senior developers will evolve. Again, AI will not displace senior developers from code review but augment their capabilities, allowing them to move from resolving petty issues to management.
Focus on Decision Making
Senior developers can now focus on making high-level decisions that guide the project instead of getting bogged down with small tasks. AI will handle the small tasks. This is a better use of expertise.
Focus on Complex Concerns and Project Management.
There are complex concerns in a software development cycle that need the attention of senior developers. Scalable pipeline deployment and project workflow management are some examples. When AI handles routine code review tasks, senior developers can focus on more important matters that impact the overall software.
AI Tool Management
Senior developers can use their expertise to select the right AI tool for their team and integrate it into their existing workflows. Since they understand their project requirements better, they can configure the AI tool to enhance team productivity. Think of it as managing the tool that does your work for you and only stepping in (to correct or reconfigure the AI) when you think it is not doing it as it should.
More Mentorship
Manually doing all the tasks affects the time a senior dedicates to mentoring juniors. With that out of the way (due to AI automation), mentorship can take priority. Helping juniors master their craft and become well-rounded increases the quality of their contribution to your team. Furthermore, seniors can also help juniors interpret AI suggestions, aligning with team goals.
Stakeholder Alignment
Aligning project goals and expectations with stakeholders helps a business run smoothly. Stakeholder alignment requires effective communication, evaluating existing processes, finding common ground, mapping objectives and project deliverables, etc. Combining these with task execution may slow down progress. With smaller tasks out of the way, seniors can now focus on stakeholder alignment.
It is not like seniors have not been taking on these strategic tasks. However, AI code review will free up the task load, allowing them to contribute more to those tasks.
Conclusion
In short, AI will not replace human reviewers, nor is it intended to do so. It will augment human reviews, allowing code authors and reviewers to overcome the limitations of traditional (manual) reviews by automating tasks such as syntax checks, error detection, style checks, PR analysis, etc.
AI will enable senior developers to focus on strategic management by automating redundant tasks, thereby boosting overall team productivity.
Finally, you should use AI as an augmented code reviewer. A human engineer should always have the final review and approval for any PR.
If you have any questions about the code reviews or about engineering, startups, marketing, or business in general, please find me on Twitter: @TheAnkurTyagi. I’d be more than happy to discuss them.
